AWS Security ChangesHomeSearch

AWS waf: New use case: detecting bots behind residential proxies via token provenance

Service: waf · 2026-11-05 · Documentation medium

File: waf/latest/developerguide/waf-bot-control-use-cases.md · Type: bot-mitigation

Summary

Adds a use case describing evasive bots that route traffic through residential proxies so IP-based controls cannot separate them from real users, and explains using targeted protection behavior/browser/token analysis plus the 'high' token provenance label (token acquired from one IP category, presented from another) in a label match rule to challenge or block such requests.

Security assessment

Documents a detection technique (token provenance label) for evasive bot traffic that defeats IP-based filtering, and how to write a label match rule to challenge or block it. This is defensive security guidance for a known evasion pattern, not a fix for a specific vulnerability or CVE.

Evidence

+Evasive bots route their traffic through residential proxy networks, so their requests arrive from the same internet providers and address ranges as your real users.

Diff

diff --git a/waf/latest/developerguide/waf-bot-control-use-cases.md b/waf/latest/developerguide/waf-bot-control-use-cases.md
index 0078753..4c25951 100644
--- a//waf/latest/developerguide/waf-bot-control-use-cases.md
+++ b//waf/latest/developerguide/waf-bot-control-use-cases.md
@@ -47,0 +48,4 @@ Scrapers, crawlers, and automated tools can generate high traffic volume, increa
+###### Detecting bots that hide behind residential proxies
+
+Evasive bots route their traffic through residential proxy networks, so their requests arrive from the same internet providers and address ranges as your real users. Controls that are based on the IP address can't separate the two without also blocking real users. Targeted protection evaluates these clients on their behavior, browser characteristics, and token rather than on their address. It also labels one pattern specifically: a token that was acquired from one category of IP address and presented from another. For example, a bot solves a challenge from a data center and then uses the resulting token from residential addresses. Legitimate clients rarely make that change. To challenge or block these requests, write a label match rule that uses the `high` token provenance label, alone or combined with other Bot Control signals. For an example, see [Example – Challenging clients whose token came from a different type of network](./waf-bot-control-example-token-provenance.html).
+